A node is a computer running Ethereum's software, and that software is called a client. Thousands of independent nodes each keep their own copy of Ethereum and check each other, which is what makes the network trustless. Since the Merge, a full node runs two clients together: an execution client and a consensus client.
What is the difference between an Ethereum node and a client?
A node is the computer. A client is the software that computer runs. So a node is a machine running Ethereum's client software, joining the network to keep a copy of the chain and check it. Thousands of these nodes run around the world, owned by many different people, with no single company in charge.
Here is the easiest way to picture it. Think of a big sports game with not one referee but thousands of them, all watching the same play. For a goal to count, the referees have to agree. If one referee tries to cheat or simply gets it wrong, the rest outvote them. No single referee can decide the outcome alone. Ethereum's nodes work the same way. Each one checks every new transaction against the rules, and they only accept what the crowd agrees is valid.
This is what people mean when they call Ethereum trustless. Trustless does not mean nobody is trustworthy. It means you do not have to trust any one company, server, or person to use the network. You can rely on the fact that thousands of independent machines are all checking the same rules at once. If one node lies, the others reject it. Bitcoin secures its chain with a different design, and you can see how the two compare in Bitcoin versus Ethereum.
The word client can be confusing at first, because in everyday life a client is a customer. In software, a client is just a program that follows a shared standard. Many teams build different Ethereum clients. They are separate programs written by separate groups, but they all follow the same Ethereum rules, so they can work together on one network (source: Ethereum: Nodes and clients). That last point matters more than it sounds, and it is where the rest of this guide is headed.
Why does an Ethereum node run two clients now?
Since the Merge in September 2022, a full Ethereum node runs two clients side by side. One is the execution client, which handles transactions. The other is the consensus client, which handles agreement using proof-of-stake. They talk to each other constantly, and neither one works alone. You need both for a complete node.
It was not always this way. Before the Merge, a single program did the whole job. The Merge split Ethereum's work into two layers, so the software split too. We will not retell the Merge story here; that history lives in our companion guide on the Merge. What matters for nodes is the result: one machine, two programs, working as a pair.
A simple analogy helps. Picture a plane with two engines. One engine does not replace the other; they run together, and the plane needs both to fly safely. An Ethereum node is similar. The execution client and the consensus client each do a different job, they hand information back and forth, and the node only works when both are running.
Here is what the two layers handle, side by side:
Execution client | Consensus client | |
|---|---|---|
What it does | Processes transactions and runs the code on Ethereum | Runs proof-of-stake, the system that agrees on new blocks |
What it holds | The current state, meaning every account, balance, and contract | The list of validators and the record of what has been agreed |
Main job | Update the database when transactions run | Decide which block comes next and finalize it |
Example clients | Geth, Nethermind, Besu, Erigon, Reth | Prysm, Lighthouse, Teku, Nimbus, Lodestar |
A few terms in that table deserve a plain definition. The state is simply Ethereum's current database, the full record of who owns what right now. A validator is someone who has staked ETH to help secure the network and gets chosen to propose and check blocks. To finalize a block means the network has agreed it is permanent and will not be undone. The deeper mechanics of proof-of-stake, including how validators and finality work, are covered in our companion guide on Ethereum proof-of-stake (source: Quicknode: Ethereum clients explained).
What does the execution client do versus the consensus client?
The execution client processes transactions, runs the Ethereum Virtual Machine, and holds the current state. The consensus client runs proof-of-stake: it tracks validators, reaches agreement with other nodes, and finalizes blocks. The two pass data to each other constantly, so the node sees one coherent chain.
Start with the execution client. When you send ETH or use an app, your transaction is handled here. The execution client runs the Ethereum Virtual Machine, often shortened to the EVM, which is the engine that runs the code in smart contracts. It then updates the state, the database of balances and contract data, to reflect what just happened. If you want to understand the EVM and gas in more detail, that lives in our guide on how Ethereum works. Popular execution clients include Geth, Nethermind, Besu, Erigon, and Reth. They are built by different teams, but they all produce the same results from the same rules.
Now the consensus client. This is the half that came from proof-of-stake. It does not run your transactions. Instead, it talks to the rest of the network to agree on which block of transactions comes next and to confirm that earlier blocks are final. It keeps track of validators and makes sure the node follows the same chain everyone else does. Popular consensus clients include Prysm, Lighthouse, Teku, Nimbus, and Lodestar.
The two clients are partners. The consensus client decides which block is next; the execution client runs the transactions inside that block and updates the state. They talk to each other over a built-in local connection called the Engine API. Then they confirm the result with each other. Picture a librarian and an accountant working at one desk. The librarian, the consensus client, decides the order in which entries get added. The accountant, the execution client, does the math and updates the books. Neither can run the library alone. That pairing is why a node now needs both pieces of software, and why a problem in either one stops the whole node (source: Ethereum: Node architecture).
What are full nodes, archive nodes, and light nodes?
Ethereum has three main kinds of node. A full node downloads and checks every block and enforces the rules; it is the most common type. An archive node is a full node that also keeps all of Ethereum's old history. A light node keeps very little and relies on full nodes for data, so it can run on small devices.
The differences come down to how much each one stores and checks. A full node is the workhorse. It verifies every block for itself, so it never has to trust anyone else's word about what is valid. This is the type most people mean when they talk about running a node, and it is what keeps the network honest, because each full node independently enforces the rules.
Here is how the three compare in plain terms:
Full node. Downloads and verifies every block, keeps recent state, and enforces the rules itself. The standard choice for trustless verification.
Archive node. A full node that also stores all of Ethereum's historical state, going back to the start. It is large and heavy, so it is mainly used by developers, data services, and analysts who need old records.
Light node. Keeps almost nothing and asks full nodes for the data it needs. It uses few resources, so it can run on a phone or a small device, but it leans on full nodes rather than checking everything itself.
For most readers, the takeaway is simple. The full node is the one that does the trustless checking. Archive nodes are a heavier version for people who need history, and light nodes are a lighter version for devices that cannot store much. Whichever type you picture, remember it still runs the two-client pair from the section above.
What is client diversity, and why does it matter?
Client diversity means the network's nodes run a mix of different client software, instead of everyone using the same program. It matters because execution clients are lopsided: one client, Geth, has long been dominant, at times running a majority of execution nodes. That much concentration in any single client is a real risk. A bug in a client that big could threaten the whole chain.
This is the part most beginner guides skip, and it is the most important idea in this article. Remember that many teams build Ethereum clients, and they all follow the same rules. That variety is a safety feature, not an accident. If one client has a hidden bug, the other clients keep working correctly and outvote the broken one, so the network heals itself. But that only works if no single client is too dominant.
Here is the risk in one sentence. If one client runs a large majority of the network and that client has a serious bug, the bug becomes the majority view, and the healthy clients can be outvoted instead of the broken one. That is why Geth's dominance among execution clients worries the people who watch Ethereum closely. Its share has swung a lot: it peaked near 85 percent in early 2024, then fell sharply after large operators moved to other clients, with rivals like Nethermind gaining ground. By 2025 no single execution client held the kind of clear majority Geth once did, though execution-client concentration is still something the community watches closely. Because the number keeps shifting, check the live share of each client yourself (source: Client Diversity tracker).
The healthy goal is balance. A widely shared rule of thumb is that no single client should run more than about a third of the network, roughly 33 percent. Below that line, a single broken client cannot drag the majority with it. Above it, the danger grows (source: Ethereum: Client diversity). The consensus layer, the proof-of-stake side, is more evenly spread than the execution layer today, but it is still more concentrated than the ideal.
So how does diversity get fixed? By people choosing a minority client when they set up a node. On the execution side, that means picking something other than the dominant client, such as Nethermind, Besu, Erigon, or Reth. On the consensus side, it means spreading across Lighthouse, Teku, Nimbus, Lodestar, and others. Each person who runs a minority client makes the network a little safer for everyone, including people who never run anything at all. Bitcoin handles its node software differently, a contrast covered in Bitcoin nodes versus miners versus wallets.
Do you need to run an Ethereum node?
For most people, no. You can use Ethereum perfectly well through a wallet or an exchange without running anything yourself. You only need your own node in two cases: you want to verify the chain yourself, trusting no one else, or you want to run a validator and help secure the network with staked ETH.
Let us be honest about the cost, because it sets expectations. Running a full node needs a reasonably powerful computer with a fast SSD, a solid-state drive, of roughly 2 terabytes or more. It also needs to stay online and stay updated. This guide will not walk you through the setup step by step; if you want to try it, the official setup instructions live in the Ethereum docs (source: Ethereum: Run a node). The point here is just to be clear about who actually benefits from doing it.
So who should run one? Three groups, mainly. People who want full trustless verification, meaning they check the chain with their own machine and never rely on someone else's server. People who run a validator and stake ETH, since a validator needs its own node to take part. And the technically curious who want to support client diversity by running a minority client, as covered above. If you are none of those, using a wallet is completely normal and safe.
From Blofin's operational perspective, the fact that thousands of independent nodes each keep their own copy of Ethereum is exactly why we can confirm a deposit with confidence: we read the chain from our own nodes and cross-check, rather than trusting any single source. The node network is the reason an Ethereum balance is something we can verify, not just take on faith.
That is the quiet power of this whole system. Whether or not you ever run a node, the network only works because enough other people and services do, each checking the rules for themselves. You get the benefit of a chain that no single company controls, simply because so many independent machines are watching it at once.
Frequently asked questions
What is the difference between an Ethereum node and a client?
A node is a computer running Ethereum's software. The client is that software. Thousands of independent nodes each keep their own copy of the chain and check each other, so no single company is in charge. Since the Merge, a full node runs two clients together: an execution client and a consensus client.
What is the difference between an execution client and a consensus client?
The execution client processes transactions, runs the code on Ethereum, and holds the current state, meaning all balances and contract data. The consensus client runs proof-of-stake: it tracks validators, agrees with other nodes on new blocks, and finalizes them. Both run on a full node, they pass data to each other, and neither works alone.
What is Ethereum client diversity, and why does it matter?
Client diversity means nodes run a mix of different client software, not one dominant program. It matters because execution clients are lopsided: Geth has long been dominant, at times running a majority of execution nodes (its share peaked near 85 percent in early 2024 and has moved since, so check a live tracker). A serious bug in a client that dominant could threaten the chain. The healthy goal is no single client above about a third of the network.
What are full, archive, and light nodes?
A full node downloads and verifies every block and enforces the rules; it is the most common type and the one that keeps the network honest. An archive node is a full node that also keeps all of Ethereum's old history, so it is large and used mainly for data and analytics. A light node keeps very little and relies on full nodes, so it can run on small devices.
Do I need to run an Ethereum node to use Ethereum?
No. Most people use Ethereum through a wallet or an exchange and never run a node. You only need your own node if you want to verify the chain yourself without trusting anyone, or if you want to run a validator. Running a full node needs a reasonably powerful computer with a fast SSD of roughly 2 terabytes or more.
Researched and written by the Blofin Academy editorial team with AI-assisted drafting. Primary sources include Ethereum: Nodes and clients and Ethereum: Client diversity, with live client-share figures from Client Diversity. All facts independently verified against cited documentation current as of June 2026; client-share percentages move over time and should be checked live.
This article is for informational purposes only and does not constitute financial advice, investment guidance, or a recommendation to buy, sell, or hold any digital asset. Cryptocurrency markets involve significant risk and you should conduct your own research and consult qualified professionals before making decisions. Blofin Academy content reflects the state of public information at time of publication; network details and client figures change over time.